test

vaedecode_fp8ization_wan21_fp8_20260109_233933_0eee5653_e567f199.log

@@ -0,0 +1,209 @@
+[2026-01-09 23:39:33] [INFO] === WAN21 ONNX->ModelOpt Quantization Run START ===
+[2026-01-09 23:39:33] [INFO] log_path: /ComfyUI/output/vaedecode_fp8ization_wan21_fp8_20260109_233933_0eee5653_e567f199.log
+[2026-01-09 23:39:33] [INFO] LD_LIBRARY_PATH already contains: /usr/lib/x86_64-linux-gnu
+[2026-01-09 23:39:33] [INFO] Versions:
+{
+  "python": "3.12.3 (main, Nov  6 2025, 13:44:16) [GCC 13.3.0]",
+  "numpy": "1.26.4",
+  "onnx": "1.19.1",
+  "onnxruntime": "1.22.1",
+  "tensorrt": "10.14.1.48.post1",
+  "modelopt": "0.40.0"
+}
+[2026-01-09 23:39:33] [INFO] onnxruntime.get_available_providers(): ['AzureExecutionProvider', 'CPUExecutionProvider']
+[2026-01-09 23:39:33] [INFO] workdir: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199
+[2026-01-09 23:39:33] [INFO] Copied ONNX into workdir:
+  src: /ComfyUI/custom_nodes/vaedecode_fp8ization.onnx
+  dst: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization.onnx
+[2026-01-09 23:39:34] [INFO] No external data files were copied (model may be embedded).
+[2026-01-09 23:39:34] [INFO] Loaded ONNX: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization.onnx
+[2026-01-09 23:39:34] [INFO] Model inputs: ['latent']
+[2026-01-09 23:39:34] [INFO] Model outputs: ['video']
+[2026-01-09 23:39:34] [INFO] latent dims (as exported): ['B', 16, 11, 72, 42]
+[2026-01-09 23:39:34] [INFO] video dims  (as exported): ['B', 'Transposevideo_dim_1', 'Transposevideo_dim_2', 'Transposevideo_dim_3', 3]
+[2026-01-09 23:39:34] [INFO] WAN IO verified (or best-effort verified).
+[2026-01-09 23:39:36] [INFO] Initial onnx.checker.check_model OK
+[2026-01-09 23:39:36] [INFO] Created initializer: __wan_resize_roi_empty_f32 (empty float32 tensor)
+[2026-01-09 23:39:36] [INFO] Created initializer: __wan_resize_scales_1_1_2_2_f32 (float32 [1,1,2,2])
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_1/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_1/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_1/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_1/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_1/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_1/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_2/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_2/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_2/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_2/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_2/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_2/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_3/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_3/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_3/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_3/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_3/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_3/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_4/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_4/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_4/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_4/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_4/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_4/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_5/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_5/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_5/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_5/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_5/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_5/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_6/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_6/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_6/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_6/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_6/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_6/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_7/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_7/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_7/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_7/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_7/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_7/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_8/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_8/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_8/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_8/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_8/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_8/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_9/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_9/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_9/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_9/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_9/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_9/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.3/resample/resample.0_10/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.3_10/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.7/resample/resample.0_10/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.7_10/Reshape_2_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] [WAN FORCE 2x] /decoder/upsamples.11/resample/resample.0_10/Resize: roi=__wan_resize_roi_empty_f32, scales=__wan_resize_scales_1_1_2_2_f32, sizes omitted. inputs=['/decoder/upsamples.11_10/Reshape_output_0', '__wan_resize_roi_empty_f32', '__wan_resize_scales_1_1_2_2_f32', '']
+[2026-01-09 23:39:36] [INFO] ResizeFixStats:
+{
+  "resize_total": 33,
+  "resized_both_present_fixed": 0,
+  "resized_empty_scales_dropped": 0,
+  "resized_wan_forced_2x": 33,
+  "resized_unfixable": 0,
+  "resized_scales_only_ok": 0,
+  "resized_sizes_only_ok": 0
+}
+[2026-01-09 23:39:36] [INFO] Model opset (pre-quant): 18
+[2026-01-09 23:39:36] [WARN] Model opset 18 < 19; attempting convert_version(model, 19) for FP8.
+[2026-01-09 23:40:04] [INFO] Opset conversion to 19: OK
+[2026-01-09 23:40:04] [INFO] Saved ONNX: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx (external=True, location=vaedecode_fp8ization_wan21_fixed.onnx_data)
+[2026-01-09 23:40:05] [INFO] onnx.checker.check_model OK
+[2026-01-09 23:40:11] [INFO] ORT CPU InferenceSession load OK
+[2026-01-09 23:40:12] [INFO] Wrote calibration npz: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_calib_latent_64.npz (latent shape=(64, 16, 11, 72, 42), dtype=float32, seed=0)
+[2026-01-09 23:40:12] [WARN] Requested EP 'cuda:0' but CUDAExecutionProvider not present in ORT provider list.
+[2026-01-09 23:40:12] [WARN] Requested EP 'trt' but TensorrtExecutionProvider not present in ORT provider list.
+[2026-01-09 23:40:12] [INFO] Filtered calibration_eps: ['cpu']
+[2026-01-09 23:40:12] [INFO] ModelOpt Python API kwargs:
+{
+  "onnx_path": "/ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx",
+  "output_path": "/ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fp8_quant.onnx",
+  "quantize_mode": "fp8",
+  "calibration_method": "entropy",
+  "calibration_data": "NpzFile '/ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_calib_latent_64.npz' with keys: latent",
+  "calibration_eps": [
+    "cpu"
+  ],
+  "override_shapes": "latent:1x16x11x72x42",
+  "use_external_data_format": true,
+  "calibrate_per_node": true,
+  "dq_only": true,
+  "disable_mha_qdq": true,
+  "keep_intermediate_files": true,
+  "high_precision_dtype": "fp32"
+}
+[2026-01-09 23:40:24] [ERROR] ModelOpt Python API failed: Data of TensorProto ( tensor name: onnx::Mul_32598) should be stored in vaedecode_fp8ization_wan21_fixed_static.onnx_data, but it doesn't exist or is not accessible.
+[2026-01-09 23:40:24] [ERROR] Traceback (most recent call last):
+  File "/ComfyUI/custom_nodes/comfyui-vae-tensorrt/nodes/gpt1_onnx_to_fp8.py", line 986, in _modelopt_quantize_python_api
+    moq.quantize(**kwargs)
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 413, in quantize
+    ) = _preprocess_onnx(
+        ^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 166, in _preprocess_onnx
+    onnx_model, is_duplicated_constant = duplicate_shared_constants(
+                                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/utils.py", line 546, in duplicate_shared_constants
+    values=tensor.values,
+           ^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 391, in values
+    self._values = self._values.load()
+                   ^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 257, in load
+    return np.array(onnx.numpy_helper.to_array(self.tensor))
+                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx/numpy_helper.py", line 414, in to_array
+    onnx.external_data_helper.load_external_data_for_tensor(tensor, base_dir)
+  File "/opt/venv/lib/python3.12/site-packages/onnx/external_data_helper.py", line 53, in load_external_data_for_tensor
+    external_data_file_path = c_checker._resolve_external_data_location(  # type: ignore[attr-defined]
+                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+onnx.onnx_cpp2py_export.checker.ValidationError: Data of TensorProto ( tensor name: onnx::Mul_32598) should be stored in vaedecode_fp8ization_wan21_fixed_static.onnx_data, but it doesn't exist or is not accessible.
+
+[2026-01-09 23:40:24] [WARN] Python API not used: ModelOpt Python API failed: Data of TensorProto ( tensor name: onnx::Mul_32598) should be stored in vaedecode_fp8ization_wan21_fixed_static.onnx_data, but it doesn't exist or is not accessible.
+[2026-01-09 23:40:24] [INFO] ModelOpt CLI command:
+  /opt/venv/bin/python -m modelopt.onnx.quantization --onnx_path /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx --quantize_mode fp8 --calibration_method entropy --calibration_data_path /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_calib_latent_64.npz --calibration_eps cpu --override_shapes latent:1x16x11x72x42 --output_path /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fp8_quant.onnx --log_level INFO --log_file /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_modelopt_fp8.log --use_external_data_format --keep_intermediate_files --disable_mha_qdq --dq_only --calibrate_per_node --high_precision_dtype fp32
+[2026-01-09 23:40:40] [INFO] ModelOpt CLI returncode: 1
+[2026-01-09 23:40:40] [INFO] ModelOpt CLI stdout (tail):
+[modelopt][onnx] - INFO - Starting quantization process for model: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx
+[modelopt][onnx] - INFO - Quantization mode: fp8
+[modelopt][onnx] - INFO - Preprocessing the model /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx
+[modelopt][onnx] - INFO - Model has dynamic inputs: ['latent']
+[modelopt][onnx] - INFO - Setting 'latent' shape to [1, 16, 11, 72, 42]
+[modelopt][onnx] - INFO - Found 0 custom layers and 52764 tensors
+[modelopt][onnx] - INFO - No custom ops found. If that's not correct, please make sure that the 'tensorrt' python package is correctly installed and that the paths to 'libcudnn*.so' and TensorRT 'lib/' are in 'LD_LIBRARY_PATH'. If the custom op is not directly available as a plugin in TensorRT, please also make sure that the path to the compiled '.so' TensorRT plugin is also being given via the  '--trt_plugins' flag (requires TRT 10+).
+[modelopt][onnx] - INFO - Duplicating shared constants
+
+[2026-01-09 23:40:40] [WARN] ModelOpt CLI stderr (tail):
+Traceback (most recent call last):
+  File "<frozen runpy>", line 198, in _run_module_as_main
+  File "<frozen runpy>", line 88, in _run_code
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/__main__.py", line 305, in <module>
+    main()
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/__main__.py", line 271, in main
+    quantize(
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 413, in quantize
+    ) = _preprocess_onnx(
+        ^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 166, in _preprocess_onnx
+    onnx_model, is_duplicated_constant = duplicate_shared_constants(
+                                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/utils.py", line 546, in duplicate_shared_constants
+    values=tensor.values,
+           ^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 391, in values
+    self._values = self._values.load()
+                   ^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 257, in load
+    return np.array(onnx.numpy_helper.to_array(self.tensor))
+                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx/numpy_helper.py", line 414, in to_array
+    onnx.external_data_helper.load_external_data_for_tensor(tensor, base_dir)
+  File "/opt/venv/lib/python3.12/site-packages/onnx/external_data_helper.py", line 53, in load_external_data_for_tensor
+    external_data_file_path = c_checker._resolve_external_data_location(  # type: ignore[attr-defined]
+                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+onnx.onnx_cpp2py_export.checker.ValidationError: Data of TensorProto ( tensor name: onnx::Mul_32598) should be stored in vaedecode_fp8ization_wan21_fixed_static.onnx_data, but it doesn't exist or is not accessible.
+
+[2026-01-09 23:40:40] [ERROR] ModelOpt CLI failed (returncode=1)
+[2026-01-09 23:40:40] [ERROR] ModelOpt output tail:
+[modelopt][onnx] - INFO - Starting quantization process for model: /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx
+[modelopt][onnx] - INFO - Quantization mode: fp8
+[modelopt][onnx] - INFO - Preprocessing the model /ComfyUI/output/wan21_work_20260109_233933_0eee5653_e567f199/vaedecode_fp8ization_wan21_fixed.onnx
+[modelopt][onnx] - INFO - Model has dynamic inputs: ['latent']
+[modelopt][onnx] - INFO - Setting 'latent' shape to [1, 16, 11, 72, 42]
+[modelopt][onnx] - INFO - Found 0 custom layers and 52764 tensors
+[modelopt][onnx] - INFO - No custom ops found. If that's not correct, please make sure that the 'tensorrt' python package is correctly installed and that the paths to 'libcudnn*.so' and TensorRT 'lib/' are in 'LD_LIBRARY_PATH'. If the custom op is not directly available as a plugin in TensorRT, please also make sure that the path to the compiled '.so' TensorRT plugin is also being given via the  '--trt_plugins' flag (requires TRT 10+).
+[modelopt][onnx] - INFO - Duplicating shared constants
+
+Traceback (most recent call last):
+  File "<frozen runpy>", line 198, in _run_module_as_main
+  File "<frozen runpy>", line 88, in _run_code
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/__main__.py", line 305, in <module>
+    main()
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/__main__.py", line 271, in main
+    quantize(
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 413, in quantize
+    ) = _preprocess_onnx(
+        ^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/quantization/quantize.py", line 166, in _preprocess_onnx
+    onnx_model, is_duplicated_constant = duplicate_shared_constants(
+                                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/modelopt/onnx/utils.py", line 546, in duplicate_shared_constants
+    values=tensor.values,
+           ^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 391, in values
+    self._values = self._values.load()
+                   ^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx_graphsurgeon/ir/tensor.py", line 257, in load
+    return np.array(onnx.numpy_helper.to_array(self.tensor))
+                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/venv/lib/python3.12/site-packages/onnx/numpy_helper.py", line 414, in to_array
+    onnx.external_data_helper.load_external_data_for_tensor(tensor, base_dir)
+  File "/opt/venv/lib/python3.12/site-packages/onnx/external_data_helper.py", line 53, in load_external_data_for_tensor
+    external_data_file_path = c_checker._resolve_external_data_location(  # type: ignore[attr-defined]
+                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+onnx.onnx_cpp2py_export.checker.ValidationError: Data of TensorProto ( tensor name: onnx::Mul_32598) should be stored in vaedecode_fp8ization_wan21_fixed_static.onnx_data, but it doesn't exist or is not accessible.
+
+[2026-01-09 23:40:40] [INFO] === RUN END ===